Scholarlink

<p>This study applies machine learning-based predictive modeling to evaluate the spatial relationship between environmental variables and historical shipwreck locations in Lake Ontario. Using a sample of 44 well-documented wrecks and nine independent geographical variables—including water depth, fetch, and proximity to submerged topography—we employed MaxEnt, a presence-only modeling tool, to test whether shipwreck locations are statistically correlated with different combinations of these potential predictors. The model demonstrates good predictive performance revealing that shallower waters, areas near submerged ridges, and zones with lower fetch values are significantly associated with shipwreck occurrence. These findings challenge assumptions about the unpredictable nature of shipwreck events and demonstrate that geographical factors do play a modest, but measurable, predictable role in shipwreck presence. While survey biases and data limitations constrain definitive conclusions, the results illustrate the potential of spatial predictive modeling to enhance archaeological understanding of maritime landscapes and guide heritage management. By adapting techniques from spatial ecology, this research provides a replicable framework and emphasizes the utility of quantitative tools for underwater archaeology.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Predictive Modeling of Historical Shipwreck Locations in Lake Ontario Using Machine Learning

  • 期刊论文

  • 发表日期：

    2025年9月23日

查看全文
  • James Conolly,
  • Cassidy Robertson

摘要

This study applies machine learning-based predictive modeling to evaluate the spatial relationship between environmental variables and historical shipwreck locations in Lake Ontario. Using a sample of 44 well-documented wrecks and nine independent geographical variables—including water depth, fetch, and proximity to submerged topography—we employed MaxEnt, a presence-only modeling tool, to test whether shipwreck locations are statistically correlated with different combinations of these potential predictors. The model demonstrates good predictive performance revealing that shallower waters, areas near submerged ridges, and zones with lower fetch values are significantly associated with shipwreck occurrence. These findings challenge assumptions about the unpredictable nature of shipwreck events and demonstrate that geographical factors do play a modest, but measurable, predictable role in shipwreck presence. While survey biases and data limitations constrain definitive conclusions, the results illustrate the potential of spatial predictive modeling to enhance archaeological understanding of maritime landscapes and guide heritage management. By adapting techniques from spatial ecology, this research provides a replicable framework and emphasizes the utility of quantitative tools for underwater archaeology.